From Evidence to Evident: Decisive Cosmological Evidence for the Normal Neutrino Mass Hierarchy

By combining DESI DR2 clustering data with NuFIT oscillation results, this study demonstrates that cosmological evidence has transitioned from prior sensitivity to decisive, likelihood-dominated exclusion of the inverted neutrino mass hierarchy, strongly favoring the normal ordering and predicting an effective Majorana mass in the few-meV regime.

The Gist

The Big Question: Which Neutrino Family is Heavier?

Imagine three siblings (neutrinos) who are almost identical, but we know they have different weights. For decades, scientists have been trying to figure out their weight order. There are only two possible family trees:

1. The "Normal" Order (NH): Two siblings are very light, and one is significantly heavier. Think of it like two toddlers and one adult.
2. The "Inverted" Order (IH): Two siblings are heavy, and one is very light. Think of it like two adults and one toddler.

For a long time, we couldn't tell which family tree was correct. Oscillation experiments (which measure how these particles change as they fly) told us the differences in their weights, but not their absolute weights. It was like knowing the age gap between siblings but not knowing if they were all 5 years old or all 50.

The New Clue: The Cosmic Scale

This paper argues that we finally have enough data to solve the mystery, thanks to a new "cosmic scale" provided by the DESI (Dark Energy Spectroscopic Instrument) and the Planck satellite.

The Analogy:
Imagine you are trying to guess the total weight of a backpack.

• The "Normal" backpack has a minimum weight of 59 grams (because the lightest sibling must weigh at least a tiny bit).
• The "Inverted" backpack has a minimum weight of 99 grams (because two siblings must be heavy).

For years, our cosmic scale was imprecise. It said, "The backpack weighs less than 200 grams." This didn't help much, because both 59g and 99g fit comfortably under 200g.

The Breakthrough:
The new DESI data is incredibly precise. It says, "The backpack weighs less than 64 grams."

Suddenly, the math changes drastically:

• The Normal backpack (min 59g) fits perfectly under 64g.
• The Inverted backpack (min 99g) is impossible. It is too heavy to fit the measurement.

The paper calculates that the "Inverted" option is now so unlikely that it is effectively ruled out by the data. The odds are so heavily stacked against it that the authors call the evidence "decisive."

The Debate: Is the Result Biased?

In science, there is always a worry: "Did we just pick the answer we wanted because of how we set up the math?"

The authors were very careful to test this. They used two different "rulers" (mathematical priors) to measure the probabilities:

1. Ruler A (SJPV): Assumes the three neutrinos come from a shared "family recipe" where their masses are related.
2. Ruler B (HS): A very neutral, "objective" ruler that doesn't assume any family relationship.

The Result:
Even with these two very different rulers, the result was the same. Both rulers pointed to the Normal order.

• The paper shows that the result isn't a trick of the math; it's because the data (the 64g limit) is so tight that it physically squeezes out the "Inverted" option.
• They even tested different "measuring tapes" (linear vs. logarithmic scales) and found that while the exact numbers changed slightly, the conclusion never wavered. The "Normal" order wins no matter how you look at it.

What This Means for Future Experiments

The paper connects this cosmic discovery to a specific type of experiment on Earth called neutrinoless double-beta decay.

• The Old Hope: If the "Inverted" order were true, these experiments would easily detect a signal. Scientists were building massive detectors expecting to find it.
• The New Reality: Since the "Normal" order is now the likely winner, the signal these experiments are looking for is expected to be extremely faint—so faint that it might be invisible even to the most sensitive future machines.

The Analogy:
Imagine you were building a net to catch a specific type of fish that you thought was swimming in the shallow water (Inverted Order). Now, the cosmic data tells you the fish is actually swimming in the deep, dark ocean (Normal Order).

• The paper predicts that your shallow-water net will likely come up empty.
• If you do catch a fish in the shallow water, it would mean either your cosmic scale was wrong, or there is some completely new, unexpected physics happening that we haven't thought of yet.

Summary

1. The Problem: We didn't know if neutrinos were "Light-Light-Heavy" or "Heavy-Heavy-Light."
2. The Evidence: New, ultra-precise cosmic data says the total weight of neutrinos is very low.
3. The Conclusion: The "Heavy-Heavy-Light" (Inverted) option is too heavy to fit the data. The "Light-Light-Heavy" (Normal) option is the only one that fits.
4. The Certainty: This conclusion is robust. It holds up even when you change the mathematical rules or assumptions.
5. The Impact: Future experiments looking for a strong signal from the "Heavy" neutrinos will likely find nothing, because the "Heavy" scenario is effectively ruled out by the universe itself.

Technical Summary

Technical Summary: From Evidence to Evident: Decisive Cosmological Evidence for the Normal Neutrino Mass Hierarchy

Problem Statement
The neutrino mass ordering (Normal Hierarchy, NH, vs. Inverted Hierarchy, IH) remains a central unresolved question at the intersection of particle physics and cosmology. While oscillation experiments have established non-zero mass splittings, they do not determine the absolute mass scale or the ordering of the eigenstates. Oscillation data define two distinct "filaments" in mass space with irreducible lower limits on the sum of neutrino masses ($\Sigma m_\nu$): $\Sigma m_\nu \simeq 0.059$ eV for NH and $\Sigma m_\nu \simeq 0.099$ eV for IH.

Historically, cosmological constraints on $\Sigma m_\nu$ were too loose to decisively distinguish between these two scenarios, leading to Bayesian evidence that was either weak or highly sensitive to the choice of prior. The central problem addressed is whether current cosmological data have reached a precision regime where the preference for NH becomes a likelihood-dominated conclusion rather than a prior-dependent artifact.

Methodology
The authors perform a Bayesian model comparison between the NH and IH hypotheses using the following framework:

1. Data Combination:

   • Cosmology: The analysis utilizes the DESI Data Release 2 (DR2) Baryon Acoustic Oscillation (BAO) clustering analysis combined with Planck CamSpec CMB data within the baseline flat $\Lambda$CDM model. The marginalized posterior for $\Sigma m_\nu$ is modeled as a truncated Gaussian with a 95% confidence upper limit of $\Sigma m_\nu < 0.0642$ eV. This limit is noted to be close to the NH floor but well below the IH minimum.
   • Oscillations: Constraints on mass splittings ($\Delta m^2_{21}, \Delta m^2_{3\ell}$) and mixing angles are taken from the latest NuFIT global analysis (incorporating JUNO reactor results), which shows a slight preference for NH ($\Delta \chi^2 = 9.4$).

2. Bayesian Evidence Calculation:

   • The Bayes factor $K = P(D|NH)/P(D|IH)$ is computed by integrating the likelihood over the prior parameter space for each hierarchy.
   • Prior Robustness: To address historical debates regarding prior sensitivity, the authors evaluate the evidence across a "prior design space" defined by two axes:
     • Measure: Linear (location parameter) vs. Logarithmic (scale-invariant).
     • Structure: Hierarchical (exchangeable masses drawn from a common parent distribution) vs. Non-hierarchical (fixed density).
   • This approach constructs four distinct priors:
     1. SJPV: Hierarchical + Logarithmic (physically motivated).
     2. HS: Non-hierarchical + Linear (reference prior, minimally informative).
     3. Hierarchical + Linear: A new construction.
     4. Non-hierarchical + Logarithmic: A new construction.
   • The analysis also tests robustness against the $w_0w_a$CDM extension (time-varying dark energy).

3. Derived Observables:

   • Posterior distributions for individual mass eigenstates ($m_1, m_2, m_3$) and the effective Majorana mass ($m_{\beta\beta}$) are derived via Monte Carlo sampling, propagating uncertainties in mixing angles and Majorana phases.

Key Results

1. Decisive Evidence for Normal Hierarchy:

   • Under the baseline $\Lambda$CDM model, the Bayes factor $K$ exceeds 460 even for the conservative HS reference prior and reaches $\sim 10^4$ for the SJPV prior.
   • This places the evidence in the "decisive" regime (Jeffreys scale $K > 100$).
   • The conclusion is driven by the fact that the entire kinematically allowed region for IH ($\Sigma m_\nu \geq 0.099$ eV) lies in the tail of the cosmological likelihood, which is centered near the NH floor.

2. Robustness to Prior Choice:

   • The preference for NH is robust across all four prior constructions in the 2x2 design space.
   • Measure vs. Structure: The residual prior dependence is governed almost entirely by the choice of measure (logarithmic vs. linear), which shifts $\ln K$ by $\sim 2.34$ (a factor of $\sim 10$). The choice of structure (hierarchical vs. non-hierarchical) has a much smaller effect ($\Delta \ln K \approx 0.75$, a factor of $\sim 2$).
   • The logarithmic measure inherently favors the NH because the middle mass eigenstate in NH ($m_2 \approx 8.6$ meV) is significantly lighter than in IH ($m_2 \approx 50$ meV), and a $1/m$ measure assigns higher prior density to lower masses.

3. Prior-Family Odds:

   • When treating the SJPV and HS families as competing models with equal prior probability, the data favor the SJPV family over the HS family by a Bayes factor of $\sim 4,700$. This suggests the data prefer the scale-adaptive, exchangeable prior structure.

4. Implications for Effective Majorana Mass ($m_{\beta\beta}$):

   • The favored NH scenario predicts an effective Majorana mass in the few-meV regime: median $m_{\beta\beta} = 3.28$ meV with a 95% credible interval of $0.95 < m_{\beta\beta} < 11.55$ meV.
   • This is significantly below the canonical IH target band ($18–50$ meV). Consequently, the next generation of neutrinoless double-beta decay experiments (e.g., LEGEND-1000, CUPID) are predicted to have only a 15–20% probability of detection under the NH hypothesis.

5. Model Robustness:

   • In the $w_0w_a$CDM extension, the constraint on $\Sigma m_\nu$ relaxes slightly, reducing the Bayes factor. However, the evidence remains in the "strong" to "decisive" regime ($K > 40$ for HS, $K \gtrsim 100$ for SJPV), indicating the result is not contingent on the strictest $\Lambda$CDM assumptions.

Significance and Claims
The paper claims to mark a qualitative transition in neutrino mass hierarchy determination:

• From Prior-Sensitive to Likelihood-Dominated: Unlike previous eras where evidence was driven by geometric prior volume penalties (favoring NH because IH occupied a smaller volume), the current DESI DR2 constraints push the IH minimum mass into the tail of the likelihood itself. The exclusion of IH is now driven by the incompatibility between the IH mass floor and the observed cosmological data, rather than just prior assumptions.
• Resolution of Historical Debates: The analysis clarifies that previous disagreements between different prior choices were due to the data not yet being precise enough to break the prior-driven degeneracy. The DESI DR2 precision has broken this degeneracy.
• Experimental Impact: The results imply that a null result in upcoming neutrinoless double-beta decay experiments is the natural outcome of the standard cosmological inference, whereas a positive signal in the IH band would require non-standard physics or a revision of the cosmological interpretation.

The authors conclude that the combination of precision cosmology, oscillation data, and transparent prior construction has rendered the neutrino mass ordering a "quantitatively decisive question" rather than a suggestive one.